A figure with a pedigree
Observed, declared, published or estimated, labelled on the page, with a range wherever it was estimated.
Working together
An instrument, not an invoice for hours
Every engagement leaves behind a record of a decision that someone else can check, and a baseline you own.
I don't sell software licences or advisory hours. I deliver instruments: defined inputs, a documented method, a checkable output and a stated limit. The second run costs a fraction of the first, and comparing the two is where the value lands.
I What you receive
Four things come out of every instrument, whatever its size.
A stated limit
What the instrument deliberately does not measure or decide. An instrument that claims to answer everything answers nothing reliably.
A named decision
The decision and its owner. I name it; I don't take it.
A review date
The date the recommendation can be judged against what actually happened. It's what makes the second run worth having.
| Not this | Nor this | This |
|---|---|---|
| A software licence. I resell no third-party platform and take no referral commission, so a readiness assessment that concludes “no new system needed” is a valid outcome, and it costs me nothing to say so. | Advisory hours: an engagement that leaves nothing behind but a document and a relationship you have to buy again next year. | An instrument: run again, by anyone, against the same definitions. Results, definitions and versions stay with you, whether or not there is a next engagement. |
II Where to start
Start with the question that's actually bothering you. Each of these is a complete engagement: you can stop after any one and keep everything you've received.
- “We keep losing people we just hired.”An offer audit on the last twelve months, then the Employer Report on live offers. No platform, no integration.
- “Everyone is busy and nothing moves.”The Organisational Debt Diagnostic. The output is a costed register with a disposition for every item.
- “An audit asked how a decision was reached, and we couldn't say.”A governance review and the Decision Canvas on the decisions that matter most.
- “Staff are asking about pay, and we don't have an answer.”Compensation benchmarking first, then a pay equity review if the gap is real.
- “We're thinking about decision software.”An analytics readiness assessment: what your data and systems can support, before you spend.
- “We want to try this on one decision first.”A single Employer Report on a live offer. Same day, no commitment, and it will tell you whether the rest is worth your time.
Prices come from one published rate card and are shared in a proposal, never invented inside one.
III Who decides what
This table makes the human-in-the-loop boundary explicit. It's the one most often blurred by a supplier who wants to be indispensable.
| Matter | Decided by | My role |
|---|---|---|
| What the organisation does about a finding | You | Name the decision, cost the options, state the trade-off |
| Whose interest the design serves where interests conflict | You, at named-executive level | Surface the conflict rather than settle it in a template |
| Whether a figure is fit to publish internally | You | Mark its pedigree and range so you can judge |
| Whether the method was followed | Either party, against the standard | Publish the standard and the version stamps |
| Whether a figure is arithmetically correct | Me, and my error is mine to fix at my cost | Re-derive it on request |
| Whether the work fits the delivery window | Me, before award, in writing | Decline or reschedule rather than absorb |
| A legal, tax, audit or actuarial conclusion | A registered professional | Name who should give it. I don't. |
The one thing I ask you to hold: a single named decision owner with actual authority. Not a committee, not a liaison. An engagement with no owner produces a report instead of a decision.
IV An engagement, end to end
A five-week diagnostic, as it actually runs. Shorter instruments compress this; none of them skip a stage.
- Scope. The decision is named and its owner identified. Scope and exclusions are fixed in writing. Staff and, where present, organised labour are briefed on what's being measured and what it will never be used for, before any data is touched.
- Measure. Anonymous time sampling, system extracts and structured interviews. Sampling is voluntary and reported only for groups of five or more. Where a number can't be measured, I say so and estimate it with a range.
- Compute and check. One results file. Figures calculated on your own tables, each marked with its pedigree, with method and engine versions stamped. Review is against the documented method, not a colleague's opinion.
- Render and hand over. Views for finance, operations, technology, employee relations, organised labour, staff and the executive, all from the one file. Different weightings, identical numbers.
- Review. Data is returned or destroyed on schedule. The baseline, definitions and versions stay with you. On the review date the instrument is re-run, and the comparison is the point.
What I need from you
A named decision owner with authority. Access to the relevant extracts. Roughly six to ten hours of your people's time across the engagement. Permission to brief staff honestly about what's being measured.
What you won't be asked for
A data warehouse, a platform purchase or a full population census. If your data isn't ready, a readiness assessment says so, and “not yet worth building” is an outcome I'm paid to deliver.
V What I stand behind, and what I don't
The right-hand column is the one that matters. A commitment without stated exclusions is a sales line that gets disclaimed in an annexure nobody read.
I stand behind
- The arithmetic. Every figure can be re-derived from the inputs at the stamped versions.
- That the method was followed, item by item, and anyone can check it.
- That the pedigree is honest. Nothing estimated is shown as observed, and thin sources are labelled thin.
- That it reproduces. Same inputs and versions give the same answer, or it's a defect.
- That the data is handled as agreed. Minimised, reported in groups of five or more, returned or destroyed on schedule.
I don't stand behind
- Any commercial outcome. A finding doesn't guarantee a saving. What happens next is a decision I name but don't take.
- Inputs I didn't measure. A figure you supply is marked declared: I warrant my use of it, not its accuracy.
- Individual behaviour. An estimate is a probability over a population, never evidence about a named person.
- Legal, tax, audit or actuarial conclusions, or compliance itself. The work supports King IV and the employment statutes; it doesn't discharge them.
This is the spirit of the Decision Warranty: it warrants that the process was followed. It doesn't promise an outcome, and it isn't insurance.
VI Commitments you can hold me to
Every client-facing output follows the Transparent Solution Standard. These eight commitments are testable by you, which is what makes them different from a values statement.
One result, many views
No stakeholder receives a number the others can't reproduce.
Every figure shows its source
Observed, declared, published or estimated, with ranges where estimated.
Every output shows its version
Method, engine and parameter versions are printed, so a result can be re-run and checked.
Limits are stated
Each output says what it deliberately doesn't measure or decide.
People decide
I name the decision and its owner. No instrument of mine issues an automated rejection of a person.
Systems before people
Organisational diagnostics never measure, rank or name individuals.
Personal information is minimised
Only what a decision needs, under POPIA. Specimens are stripped.
Outcomes are checked
Every significant recommendation carries a review date.
VII Where it applies
The method doesn't change by sector. What changes is which decision is worth instrumenting first.
| Sector | The decision worth instrumenting first | Starting instrument |
|---|---|---|
| Municipal and public entities | Appointments and procurement recommendations that must survive an audit two years later | Governance review and Decision Canvas |
| Mining and heavy industry | Scarce-skill appointments at remote operations, where an offer that doesn't cover the real cost of living fails in month four | Offer audit, then Employer Report |
| Financial services | Remuneration decisions under committee scrutiny, and metric definitions that differ between divisions | Pay equity review |
| Manufacturing and logistics | Approval queues and undocumented exception handling that nobody owns but everybody works around | Organisational Debt Diagnostic |
| Agriculture and co-operatives | Seasonal labour and input timing decisions, taken annually with no record of what last season's decision cost | Governance review |
| Professional services and recruiters | Placement advice given on instinct, and a client asking why the last one didn't hold | Employer Report |
| Education and health | Staffing and rostering decisions where the human load is obvious to everyone and measured by nobody | Organisational Debt Diagnostic |
What I won't claim: delivered engagements in every sector above. The table shows where each instrument applies and why, not where it has already been run. A sector where I have no experience will be named as such in any proposal.
VIII People, data and protection
A diagnostic that measures how an organisation works is measuring people's working lives. That deserves a stated position, not a privacy clause in the annexures.
Systems, never individuals
Findings attach to a process, a rule, a decision or a role. Time is sampled anonymously and reported only for groups of five or more.
Minimisation under POPIA
Only what the named decision needs. Data is returned or destroyed on schedule, and client personal information stays in South Africa.
Prohibited uses, in the contract
Diagnostic data may not be used in disciplinary, incapacity, appraisal or retrenchment-selection decisions. The clause survives the engagement.
Where labour is organised
A joint governance body with audit rights is recommended, and staff receive their own plain-language view of the findings.
| Obligation | Position |
|---|---|
| POPIA | Personal information minimised by design; group-of-five reporting floor; stripped specimens; a stated return-or-destroy schedule; a registered information officer. |
| Employment Equity Act | Any psychometric input must be valid, reliable, fairly applied and unbiased. No instrument triggers an automated rejection of a candidate. |
| Labour Relations Act | Diagnostic data is contractually barred from disciplinary, incapacity, appraisal and retrenchment-selection use. |
| King IV | Governance instruments support King IV obligations. They don't discharge them, and every relevant output says so. |
| Conflict of interest | No third-party software resold and no referral commission taken. Declared in every proposal. |
| Data residency | Client data is processed within the agreed jurisdiction. That's an execution guarantee, kept separate from the governance guarantee that a number means what it says. |
| Professional standing | I'm not a lawyer, auditor, actuary or tax practitioner. Where one is needed, I name the registered professional who will give the opinion. |
Every output prints its method version, parameter set and the pedigree of each figure, so it can be audited after the engagement by someone who was never in the room: an internal auditor, a commissioner, or whoever comes after me.
IX Before you start, and after
How to prepare
- Name one decision or one pain. “Our senior hires leave” is enough.
- Identify the owner who holds authority for the decision.
- Gather what exists. Imperfect data is fine; its quality gets labelled.
- Involve the people affected early.
- Agree the review date now.
Getting the most from it
- Read the ranges, not just the headline.
- Act on the named decisions. Unowned findings change nothing.
- Share the stakeholder views.
- Re-run it. The second run is where improvement becomes provable.
- Build your own capability with the guides and certification.
Questions worth asking me
- Which figures are observed, which declared, which estimated?
- What would change the recommendation?
- What does this instrument deliberately not measure?
- Who can see our data, for what purpose, and what happens to it afterwards?
- How will we know in six months whether acting on this paid off?
X Partners and associates
Because the method is documented rather than personal, others can deliver it. That's the intent: a standard only its author can apply isn't a standard.
- Accredited associateAn independent practitioner who delivers facilitation, interviews and workshops to the documented method, under review. No associate signs a finding; the company does.
- Implementation partnerA firm accredited to deliver named instruments to its own clients under licence, with certified staff, an annual quality review and the eight commitments as binding terms.
- Referral and joint bidFor tenders where decision analysis is one component of a larger scope. I'm content to be a named subcontractor, with scope and capacity stated separately.
- Academic and standards workCo-developed curriculum and joint submissions to standards and professional bodies. The method is published rather than protected, so collaboration is welcome.
A decision you can't explain is a decision you can't defend, improve or repeat.
Bring the one that's costing you now. A first conversation is about that decision, not a sale.